¹ Newstead has a little samian ware divisible between Antonine I and II (PSAS lxxxiv (1949-1950), 28 ff.)Google Scholar. Carzield, if there is only an Antonine I occupation, could be vital, but the evidence is not conclusive (Trans. Dumfriesshire and Galloway N.H. and Ant. Soc., 3rd ser., xxiv (1945-1946), 3 ff.). Birrens will prove the greatest value, with its epigraphic dating for the division between Antonine I and II, when the pottery from the recent work is published. Corbridge, which ought to be our most potent source for dividing Antonine I and II samian, is beset by problems from the point of view of the samian ware (see Appendix II).Google Scholar

² So with the reoccupation of Hadrian's Wall in Period Ib (RIB 1389) and so with Ilkley (RIB 639). Brough-on-Noe (RIB 283) is little help, in view of the paucity of samian ware from the Antonine levels.

³ PSAS lxv (1930-1931), 432 ff. and lxxi (1936-7), 386 f.Google Scholar

⁴ See p. 6 ff.

⁵ Unfortunately some of these may have to be changed, before the catalogue goes to press, for the sake of consistency, but the relevant museum numbers will be quoted to avoid confusion. Stamps from broken dies are noted in the form la' .OFFIC.

⁶ Form 27g is used for Dragendorff form 27 when it has a distinct groove around the footring (normally a first-century characteristic, but adopted by a few potters from Central Gaul and Chemery-Faulquemont in the early second century). Form 33a is used for varieties of Dragendorff form 33 with internal mouldings at the junction of base and wall and, usually, external grooves at the top and bottom of the wall. Forms 18R, 18/31R and 31R are those forms approximately equivalent to Dragendorff 18, 18/31 and 31 which have a rouletted ring on the base and other differences, too. This useful distinction and classification was first used by Professor Birley in dealing with dishes of form 18/31R from the Birdoswald Alley (CW ² xxx (1930), 184). James Curie had already noted the distinction and form 18/31R is his form 6 (Newstead, pl. XXXIX). The number of examples present is indicated in brackets thus: (two), (sixteen), etc.Google Scholar

⁷ Namely the Elliot, Martin and Mason Collections from Newstead.

⁸ The late Sir Ian Richmond kindly put the Inchtuthil samian in the writer's hands for study.

⁹ Tacitus, Agricola 25, with the evidence of the coins (Tacitus, Agr. ed. Ogilvie and Richmond (1967), 70.

¹⁰ PSAS lxxiii (1938-1939), 142 f.Google Scholar

¹¹ Trans. Glasgow Arch. Soc. NS 14 (1956), 54.Google Scholar

¹² Stamps of Bassus, who was primarily a pre-Flavian potter, have been noted from the following other sites first occupied under the Flavians: Butzbach, Catterick, Corbridge, Friedberg, Malton and Nijmegen (Ulpia Noviomagus).

¹³ Oswald, F., Index of Potters' stamps on Terra Sigillata (1931), 11, assigns this to G. Albinus. Gaulish potters' stamps do sometimes have only praenomen and cognomen, but here there is no stop or space between G and A and it seems best to accept a single, peregrine name, with Birley (R. E. M. Wheeler, London in Roman Times (1930), 172 No. 94).

¹⁴ PSAS lxv (1930-1931), 433.Google Scholar

¹⁵ JRS iv (1914), 27 ff.Google Scholar

¹⁶ PSAS lxv (1930-1931), 446.Google Scholar

¹⁷ Macdonald, G., The Roman Wall in Scotland, 2nd edn. (1934), 460.

¹⁸ Der Römische Limes in Österreich, xvii (1933), Abb. 24.Google Scholar

¹⁹ Stanfield, J. A. and Simpson, Grace, Central Gaulish Potters (1958), pl. 120, 4.

²⁰ Oswald, op. cit., 52, ‘Domitian-Trajan’.

²¹ Jahrbuch für Altertumskunde, vi (1913), 182.Google Scholar

²² Oswald, op. cit., 55.

²³ VCH Bedfordshire, ii (1908), 4.Google Scholar

²⁴ Wilson, D., The Archaeology and Prehistoric Annals of Scotland (1863), ii, 76.Google Scholar

²⁵ Op. cit., 96.

²⁶ Arch. Ael., 4th ser., xxviii (1950), 197.Google Scholar

²⁷ Henslow, J. S., An Account of the Roman Antiquities found at Rougham (1843), 10.

²⁸ PSAS xxxix (1904-1905), 492.Google Scholar

²⁹ Terrisse, J. R., Les céramiques sigillés gallo-romaines des Martres-de-Veyre (Gallia supplement xix, 1969), pl. LIV.Google Scholar

³⁰ Macdonald, op. cit., 250.

³¹ Wilson, op. cit., 76.

³² Many of the stamps of Iullinus of La Graufesenque were unaccountably assigned to Lezoux by Oswald (op. cit., 152).

³³ Oswald, op. cit., 102.

³⁴ Curle, Newstead, pl. XLI, 14; p. 207, Nos. 1 and 3; p. 209, 2; p. 215, 9; pl. XLIII, bottom and p. 217, 6.

³⁵ Ibid., 213, No. 6.

³⁶ Stanfield and Simpson, op. cit., (note 19), pl. 44, 513.

³⁷ PSAS xxxv (1900-1901), 183.Google Scholar

³⁸ PSAS xcv (1961-1962), 216.Google Scholar

³⁹ PSAS lxxxvi (1951-1952), 97.Google Scholar

⁴⁰ Ibid., 99 f.

⁴¹ Trans. Dumfriesshire and Galloway N.H. and A.S., 3rd ser., xxxix (1962), 24–49.Google Scholar

⁴² Discovery and Excavation in Scotland 1967, 21.

⁴³ Trans. Dumfriesshire and Galloway N.H. and A.S., 3rd ser., xli (1964), 145.Google Scholar

⁴⁴ Robertson, Anne S., The Roman Fort at Castledykes (1964), 175 ff., for Dr. Grace Simpson's report.

⁴⁵ Ibid., fig. 35A, 9.

⁴⁶ JRS lvi (1966), 270 for Professor Frere's suggestion of a single structural period.Google Scholar

⁴⁷ Trans. Dumfriesshire and Galloway N.H. and A.S., 3rd ser., xxxiv (1955-1956), 19, where ‘X-3’ should presumably read ‘X-2’ or, more probably, ‘the Potter of the Rosette’.Google Scholar

⁴⁸ Ibid., xxx (1951-52), 14.

⁴⁹ See p. 25 f. for discussion of the standard overlap for sites held throughout the Antonine period.

⁵⁰ There now seems to be no evidence for pre-Agricolan structures at Milton.

⁵¹ Tacitus, Agricola 23.

⁵² All were apparently collected at the time of the construction of the railway.

⁵³ Agricola 25 records a fort or forts (the texts vary) attacked by the enemy in the sixth season, but the scene is not necessarily north of the Isthmus.

⁵⁴ Ibid. 22 f.

⁵⁵ It is tempting to wonder whether the gates of his own town were ever in his mind (cf. Brogan, O., Roman Gaul, 1953, p. 70).

⁵⁶ Pen Llystyn could well have been replaced by a fortlet now, particularly in view of the samian ware (Arch. Journ. cxxv (1969), 101 ff.). It is perhaps not impossible that the fortlet at Castleshaw, replacing the cohort fort, could belong to the 80s, though it was certainly in use in the early years of the second century. At Bainbridge the Agricolan fort was replaced by another fort in the Flavian-Trajanic period. It is not known, however, whether there was a gap in the occupation. So little is known of the structures of most of the Flavian forts in the north of England in general that it is difficult to assess the situation. Note, however, that there is no evidence that any of the forts in the Lake District were held under Agricola.Google Scholar

⁵⁷ See p. 4, note 9.

⁵⁸ It is interesting to note its presence on sites on Hadrian's Wall. Decorated bowls and, or, late South Gaulish potters' stamps have been recorded from Wallsend, Benwell, Haltonchesters, Chesters, Birdoswald (though some could theoretically perhaps be earlier than the fort) and Stanwix. A bowl (with rivets) from Turret 48A should be noted (CW ² xxvi (1926), 441)Google Scholar. Perhaps the South Gaulish bowls from South Shields (at least three) should also be added, since Mr. Gillam assures me that there is no pre-Hadrianic occupation of the site of the known fort, though a nearby Trajanic one would not be an impossibility, especially now that the discovery of the next fort east of Corbridge in the Stanegate series suggests a system extending to the south side of the Tyne estuary (cf. Northern History i (1966), 15).Google Scholar

⁵⁹ JRS ix (1919), 126.Google Scholar

⁶⁰ See also p. 10, note 46.

⁶¹ JRS xxvii (1937), 93; xxviii (1938), 141 and xxix (1939), 5.Google Scholar

⁶² p. 11 (Dalswinton).

⁶³ p. 9 (Newstead).

⁶⁴ p. 8 (Newstead).

⁶⁵ p. 13, note 58.

⁶⁶ Frere, S. S., Britannia (1967), 122.

⁶⁷ Tacitus, , historiae I, 2.Google Scholar

⁶⁸ Syme, R., Tacitus, p. 118.

⁶⁹ Not the same potter, it seems, as Cettus of Les Martres-de-Veyre, and probably working at Lezoux.

⁷⁰ This sherd joins with pieces found previously (PSAS xciv (1960-1961), 102, No. 3), now in the NMA.Google Scholar

⁷¹ The famous stamp ‘NOCTVRNA’ really cannot be accepted on the evidence available (Macdonald, Roman Wall, 2nd edn, p. 216). It may be noted that the other supposed Westerndorf stamp (Ibid.), from Newstead, is really one of Carantinus of Lezoux.

⁷² PSAS lxv (1930-1931), 433 ff.; Oswald, op. cit., p. 265.Google Scholar

⁷³ Appendix I, p. 43.

⁷⁴ PSAS xcvii (1963-1964), 186.Google Scholar

⁷⁵ See Appendix I, p. 43.

⁷⁶ Miller, S. N., The Roman Fort at Balmuildy (1922), pl. xxxiv, 56.Google Scholar

⁷⁷ This obscure stamp is only known otherwise from Tongres (Schaetzen, de and Vanderhoeven, , Terra Sigillata te Tongeren 2 (1964), pl. XIII, 11).Google Scholar

⁷⁸ For Chresimus see Appendix I, p. 42f.

⁷⁹ p. 3, note 5.

⁸⁰ Rev. Arch, du Centre ii (1963), 48 ff. for moulds of Cinnamus at Vichy in styles different from any known at Lezoux.Google Scholar

⁸¹ E.g. T†.OFFIC is from a broken die of Ruttus, originally giving stamps reading RVT†.OFFIC; and SVRIVS is from a broken die of Casurius.

⁸² Cf. Latomus xxviii (1969), 469. Many other examples could be cited.Google Scholar

⁸³ Much of the material from Mr. J. A. Petch's excavations at Benwell cannot now be found, nor have the Newcastle stamps noted in CIL been located.

⁸⁴ RIB 283.

⁸⁵ Excluding stamps from the Wroxeter Gutter.

⁸⁶ Corbridge has a much larger collection than the other sites, and the overlap with them is correspondingly greater.

⁸⁷ p. 26.

⁸⁸ Atkinson, D., Report of Excavation at Wroxeter, 1923-1927 (1942), p. 127 ff.Google Scholar

⁸⁹ Proc. Soc. Ant. Lond. 2nd ser. xxi (1906-1907), 268 ff.; xxii (1907-9), 395 ff. A more complete list is to be found in Ephemeris Epigraphica ix, 675 ff. Others not listed there have been seen, but some stamps recorded by Oswald as from Pudding Pan Rock have been rejected, since they come from other finds off the Kent coast.Google Scholar

⁹⁰ The Wroxeter examples are all relatively shallow, with diameters and heights in the proportion of about 4 to 1. For Pudding Pan Rock the corresponding proportion is about 3 to 1.

⁹¹ The earliest date for the prototype is given by its presence in a pit at Lezoux discussed on p. 49.

⁹² For the distinction between the two varieties, Collingwood and Richmond, Archaeology of Roman Britain, 2nd edn. (1969) fig. 76. The essential point, apart from the different proportions, is that 18/31R has a slight angle between wall and base, whereas 31R has a continuous curve with only a ridge marking the junction. Form 18/31R was typical of the Birdoswald Alley (CW ² xxx (1930), 184–5, but is still present in an Alcester pit of c. 150-160 with large numbers of whole vessels consistently of that date. Its disappearance and replacement by 31R was no doubt gradual, but it is doubtful if it was made much after 160.Google Scholar

⁹³ The more interesting additions include, in the early group, the late Montans ware, several sherds by Quintilianus and X-6, and a little work of Acaunissa and Birrantus. Sissus II, Tetturo and Carantinus or Illixo are all noteworthy, but the really important additions are the bowls of Advocisus, Casurius, Mercator and Paternus II.

⁹⁴ A little-known potter of Lezoux, who used an ovolo with a tongue ending in a rosette containing a distinct cross. The latter is only seen in very clear impressions, however. His work is known from an Antonine I level at Newstead (PSAS lxxxiv (1949-1950), 27, No. 6).Google Scholar

⁹⁵ The sherd referred to in the last note.

⁹⁶ PSAS lxxii (1937-1938), 315, No. 3 and p. 299Google Scholar, taken in the light of the recent evidence (Trans. Dumfriesshire and Galloway N.H. and A.S., 3rd ser., xli (1964), 154).Google Scholar

⁹⁷ Information from Mr. G. B. Rogers, who kindly allows his important discovery to be mentioned here.

⁹⁸ Yorkshire Archaeological Journ. 26 (1920-1922), 51, No. 2. Therecent work at Slack has not produced a single sherd of Antonine samian, nor is any of the coarse pottery necessarily Antonine.Google Scholar

⁹⁹ Gallia xxvii (1969), 3 ff.Google Scholar

¹⁰⁰ Atkinson, op. cit. (note 88), pl. 35, G6, from the Wroxeter Gutter.

¹⁰¹ Gallia xxvii (1969), 6, No. 4.Google Scholar

¹⁰² Stanfield and Simpson, op. cit. (note 19), pl. 115, 6 in the style of Advocisus, but with the mouldstamp DIVIX, which does not appear in Scotland.

¹⁰³ ‘Paternus I’ is used here for the style associated with moulds and bowls with the plain-ware stamp PATERNIM in the decoration. The connexion with Ianuaris II was very close, and both their stamps were impressed on the mould of a bowl from Straubing (Walke, N., Straubing-Sorviodurum, Limesforschungen 3 (1965)) Taf. 114a). ‘Paternus II’ is used for the style appearing with the large ligatured Paternus stamp. Whether there were two men is perhaps debatable: the differences in style and date are not.Google Scholar

¹⁰⁴ Several Paternus II sherds are known from the burnt layer of the Antonine barracks at Bainbridge, for instance.

¹⁰⁵ Cassius Dio 72, 8.

¹⁰⁶ First suggested by Mr. Gillam, in Trans. Durham & Northumberland A. & A. Soc. x (1953), 375.Google Scholar

¹⁰⁷ Suggested by Dr. M. G. Jarrett in his edition of Nash-Williams, The Roman Frontier of Wales (1969), p. 20, n. 2. Dr. Mann, J. C. and Mr. Gillam, J. P. then added their similar suggestion (Arch. Ael., 4th ser., xlviii (1970), 1 ff.)Google Scholar, and finally Dr. Jarrett's, views were expressed at greater length in Bonner Jahrbücher 170 (1970), 178 ff.Google Scholar

¹⁰⁸ RIB 2148, which Dr. Mann, J. C. has suggested might belong to the reign of Commodus (Hermes 91 (1963), 487) is not conclusive, as Dr. Mann carefully points out.Google Scholar

¹⁰⁹ Trans. Durham & Northumberland A. & A. Soc. x (1953), 375, suggesting that the Antonine I evacuation came under Calpurnius Agricola.Google Scholar

¹¹⁰ Frere, S. S., Britannia, A History of Roman Britain (1967), 152.Google Scholar

¹¹¹ As well as the epigraphic evidence from Brough-on-Noe (RIB 283) and Ilkley (RIB 639), the general evidence of the pottery for the reoccupation of many others after a long gap is indisputable.

¹¹² Arch. Ael. 4th ser., xlii (1964), 25 ff.Google Scholar

¹¹³ This difficulty was noted by Frere (Britannia, 161 n. 1) but could not be resolved in 1967.

¹¹⁴ No Pennine fort has any certain evidence for a break in its Antonine occupation. Could troops have been moved from Wales? Dr. Grace Simpson has demonstrated Antonine activity at many forts (Arch. Cambrensis 1962, 103 ff.; 1963, 13 ff.), but it is by no means always certain that they were held throughout the Antonine period. Some could have been given up about 160 on the (admittedly negative and sometimes meagre) evidence of the samian ware. Sites which might tentatively be mentioned include Bryn-y-Gefeiliau, Caerhun, Cardiff and Gelligaer, but it must be stressed that there is much uncertainty. Temporary evacuation of forts for a few years could scarcely be expected to be reflected in the pottery.

¹¹⁵ RIB 1460-1.

¹¹⁶ Scriptores Historiae Augustae, M. Antoninus, 9.

¹¹⁷ Ibid.

¹¹⁸ Corbridge (RIB 1137 and 1149), Ribchester (RIB 589), Carvoran (RIB 1792 and 1809), Chesterholm (RIB 1703) and possibly Hardknot (RIB 793).

¹¹⁹ Cassius Dio 72, 8.

¹²⁰ This was always difficult for Mr. Gillam's 1953 solution. An alternative explanation, namely that it was Hadrian's Wall which was crossed, but that only a limited sector in the vicinity of Haltonchesters and Corbridge was affected, cannot be dismissed out of hand. If the Corbridge destruction could be assigned to this period, the problems of the samian ware in the deposit would be relieved a little (Appendix II). Although the acceptance of this hypothesis would have wide implications, they would not affect the arguments set out here.

¹²¹ Steer, K. A., in Richmond, I. A. (ed.) Roman and Native in North Britain (1958), 91, remains the best account.

¹²² See Appendix VII and note also the milestone from the road linking Cramond and Newstead, naming the latter, and very possibly Severan in date (RIB 2313).

¹²³ There are some sherds from another late source in South Gaul. PSAS lxxxiv (1949-1950), p. 27, No. 4 is a sherd of this class from an Antonine I context: there are also some unpublished examples from Newstead.Google Scholar

¹²⁴ Proc. Leeds Phil. & Lit. Soc. (Lit. & Hist. Section), xii (1966), p. 32 ff.Google Scholar

¹²⁵ Arch. Ael, 3rd ser., vii (1911), 165.Google Scholar

¹²⁶ Arch. Ael., 4th ser., xxviii (1950), 177.Google Scholar

¹²⁷ Ibid. 197.

¹²⁸ Ibid., xxxiii (1955), 241-2.

¹²⁹ Arch. Ael., 3rd ser., iv (1908), 254.Google Scholar

¹³⁰ Arch. Ael., 4th ser., xxxi (1953), 250.Google Scholar

¹³¹ Ibid, xxxvii (1959), 24.

¹³² Stanfield and Simpson, Central Gaulish Potters, 247 and verbal information.

¹³³ PSAS xciv (1960-1961), 113.Google Scholar

¹³⁴ Hartley, K. F. in M. R. Hull, The Roman Potters' Kilns of Colchester, p. 115.Google Scholar

¹³⁵ Curle illustrated very little in his report on Newstead.

¹³⁶ Arch. Ael., 4th ser., xlviii (1970), 34 f.Google Scholar

¹³⁷ For the last two see Hartley's, K. F. remarks in Trans. Birmingham Arch. Soc. 77 (1959), 13.Google Scholar

¹³⁸ e.g. the Inchtuthil store and the one implied at Cirencester (Antiq. Journ. xlii (1962), 5).Google Scholar

¹³⁹ Arch. Ael., 4th ser., xlviii (1970), 38.Google Scholar

¹⁴⁰ Ber. der Römanisch-Germanische Kommission 46–47 (1965-1966), 77.Google Scholar

¹⁴¹ Oelmann, F., Die Keramik des Kastells Niederbieber, p. 11, 1.

¹⁴² Archaeologiai Ertesitö 49 (1936), p. 47, 2.Google Scholar

¹⁴³ Oswald, Stamps, p. 33 is a false entry, due to misunderstanding of Oelmann's text.